1 /*
2  * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/stat.h>
18 
19 #include <errno.h>
20 #include <dirent.h>
21 #include <fcntl.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <limits.h>
26 #include <unistd.h>
27 #include <uuid.h>
28 
29 #include "got_compat.h"
30 
31 #include "got_error.h"
32 #include "got_object.h"
33 #include "got_path.h"
34 
35 #include "got_lib_fileindex.h"
36 #include "got_lib_worktree.h"
37 
38 /* got_fileindex_entry flags */
39 #define GOT_FILEIDX_F_PATH_LEN		0x00000fff
40 #define GOT_FILEIDX_F_STAGE		0x0000f000
41 #define GOT_FILEIDX_F_STAGE_SHIFT	12
42 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
43 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
44 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
45 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
46 #define GOT_FILEIDX_F_REMOVE_ON_FLUSH	0x00100000
47 #define GOT_FILEIDX_F_SKIPPED		0x00200000
48 
49 struct got_fileindex {
50 	struct got_fileindex_tree entries;
51 	int nentries; /* Does not include entries marked for removal. */
52 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
53 };
54 
55 mode_t
56 got_fileindex_entry_perms_get(struct got_fileindex_entry *ie)
57 {
58 	return ((ie->mode & GOT_FILEIDX_MODE_PERMS) >>
59 	    GOT_FILEIDX_MODE_PERMS_SHIFT);
60 }
61 
62 static void
63 fileindex_entry_perms_set(struct got_fileindex_entry *ie, mode_t mode)
64 {
65 	ie->mode &= ~GOT_FILEIDX_MODE_PERMS;
66 	ie->mode |= ((mode << GOT_FILEIDX_MODE_PERMS_SHIFT) &
67 	    GOT_FILEIDX_MODE_PERMS);
68 }
69 
70 mode_t
71 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
72 {
73 	mode_t perms = got_fileindex_entry_perms_get(ie);
74 	int type = got_fileindex_entry_filetype_get(ie);
75 	uint32_t ftype;
76 
77 	if (type == GOT_FILEIDX_MODE_REGULAR_FILE ||
78 	    type == GOT_FILEIDX_MODE_BAD_SYMLINK)
79 		ftype = S_IFREG;
80 	else
81 		ftype = S_IFLNK;
82 
83 	return (ftype | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
84 }
85 
86 const struct got_error *
87 got_fileindex_entry_update(struct got_fileindex_entry *ie,
88     int wt_fd, const char *ondisk_path, uint8_t *blob_sha1,
89     uint8_t *commit_sha1, int update_timestamps)
90 {
91 	struct stat sb;
92 
93 	if (fstatat(wt_fd, ondisk_path, &sb, AT_SYMLINK_NOFOLLOW) != 0) {
94 		if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
95 		    errno == ENOENT))
96 			return got_error_from_errno2("fstatat", ondisk_path);
97 		sb.st_mode = GOT_DEFAULT_FILE_MODE;
98 	} else {
99 		if (sb.st_mode & S_IFDIR)
100 			return got_error_set_errno(EISDIR, ondisk_path);
101 		ie->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
102 	}
103 
104 
105 	if ((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
106 		if (update_timestamps) {
107 			ie->ctime_sec = sb.st_ctim.tv_sec;
108 			ie->ctime_nsec = sb.st_ctim.tv_nsec;
109 			ie->mtime_sec = sb.st_mtim.tv_sec;
110 			ie->mtime_nsec = sb.st_mtim.tv_nsec;
111 		}
112 		ie->uid = sb.st_uid;
113 		ie->gid = sb.st_gid;
114 		ie->size = (sb.st_size & 0xffffffff);
115 		if (S_ISLNK(sb.st_mode)) {
116 			got_fileindex_entry_filetype_set(ie,
117 			    GOT_FILEIDX_MODE_SYMLINK);
118 			fileindex_entry_perms_set(ie, 0);
119 		} else {
120 			got_fileindex_entry_filetype_set(ie,
121 			    GOT_FILEIDX_MODE_REGULAR_FILE);
122 			fileindex_entry_perms_set(ie,
123 			    sb.st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
124 		}
125 	}
126 
127 	if (blob_sha1) {
128 		memcpy(ie->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
129 		ie->flags &= ~GOT_FILEIDX_F_NO_BLOB;
130 	} else
131 		ie->flags |= GOT_FILEIDX_F_NO_BLOB;
132 
133 	if (commit_sha1) {
134 		memcpy(ie->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
135 		ie->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
136 	} else
137 		ie->flags |= GOT_FILEIDX_F_NO_COMMIT;
138 
139 	return NULL;
140 }
141 
142 void
143 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *ie)
144 {
145 	ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
146 }
147 
148 void
149 got_fileindex_entry_mark_skipped(struct got_fileindex_entry *ie)
150 {
151 	ie->flags |= GOT_FILEIDX_F_SKIPPED;
152 }
153 
154 const struct got_error *
155 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
156     const char *relpath)
157 {
158 	size_t len;
159 
160 	*ie = calloc(1, sizeof(**ie));
161 	if (*ie == NULL)
162 		return got_error_from_errno("calloc");
163 
164 	(*ie)->path = strdup(relpath);
165 	if ((*ie)->path == NULL) {
166 		const struct got_error *err = got_error_from_errno("strdup");
167 		free(*ie);
168 		*ie = NULL;
169 		return err;
170 	}
171 
172 	len = strlen(relpath);
173 	if (len > GOT_FILEIDX_F_PATH_LEN)
174 		len = GOT_FILEIDX_F_PATH_LEN;
175 	(*ie)->flags |= len;
176 
177 	return NULL;
178 }
179 
180 void
181 got_fileindex_entry_free(struct got_fileindex_entry *ie)
182 {
183 	free(ie->path);
184 	free(ie);
185 }
186 
187 size_t
188 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
189 {
190 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
191 }
192 
193 uint32_t
194 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
195 {
196 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
197 }
198 
199 void
200 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
201 {
202 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
203 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
204 	    GOT_FILEIDX_F_STAGE);
205 }
206 
207 int
208 got_fileindex_entry_filetype_get(struct got_fileindex_entry *ie)
209 {
210 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
211 }
212 
213 void
214 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
215 {
216 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_ONDISK;
217 	ie->mode |= (type & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
218 }
219 
220 void
221 got_fileindex_entry_staged_filetype_set(struct got_fileindex_entry *ie, int type)
222 {
223 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_STAGED;
224 	ie->mode |= ((type << GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT) &
225 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED);
226 }
227 
228 int
229 got_fileindex_entry_staged_filetype_get(struct got_fileindex_entry *ie)
230 {
231 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_STAGED) >>
232 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT;
233 }
234 
235 int
236 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
237 {
238 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
239 }
240 
241 int
242 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
243 {
244 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
245 }
246 
247 int
248 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
249 {
250 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
251 }
252 
253 int
254 got_fileindex_entry_was_skipped(struct got_fileindex_entry *ie)
255 {
256 	return (ie->flags & GOT_FILEIDX_F_SKIPPED) != 0;
257 }
258 
259 static const struct got_error *
260 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
261 {
262 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
263 		return got_error(GOT_ERR_NO_SPACE);
264 
265 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
266 	fileindex->nentries++;
267 	return NULL;
268 }
269 
270 const struct got_error *
271 got_fileindex_entry_add(struct got_fileindex *fileindex,
272     struct got_fileindex_entry *ie)
273 {
274 	/* Flag this entry until it gets written out to disk. */
275 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
276 
277 	return add_entry(fileindex, ie);
278 }
279 
280 void
281 got_fileindex_entry_remove(struct got_fileindex *fileindex,
282     struct got_fileindex_entry *ie)
283 {
284 	/*
285 	 * Removing an entry from the RB tree immediately breaks
286 	 * in-progress iterations over file index entries.
287 	 * So flag this entry for removal and remove it once the index
288 	 * is written out to disk. Meanwhile, pretend this entry no longer
289 	 * exists if we get queried for it again before then.
290 	 */
291 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
292 	fileindex->nentries--;
293 }
294 
295 struct got_fileindex_entry *
296 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
297     size_t path_len)
298 {
299 	struct got_fileindex_entry *ie;
300 	struct got_fileindex_entry key;
301 	memset(&key, 0, sizeof(key));
302 	key.path = (char *)path;
303 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
304 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
305 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
306 		return NULL;
307 	return ie;
308 }
309 
310 const struct got_error *
311 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
312     got_fileindex_cb cb, void *cb_arg)
313 {
314 	const struct got_error *err;
315 	struct got_fileindex_entry *ie, *tmp;
316 
317 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
318 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
319 			continue;
320 		err = (*cb)(cb_arg, ie);
321 		if (err)
322 			return err;
323 	}
324 	return NULL;
325 }
326 
327 struct got_fileindex *
328 got_fileindex_alloc(void)
329 {
330 	struct got_fileindex *fileindex;
331 
332 	fileindex = calloc(1, sizeof(*fileindex));
333 	if (fileindex == NULL)
334 		return NULL;
335 
336 	RB_INIT(&fileindex->entries);
337 	return fileindex;
338 }
339 
340 void
341 got_fileindex_free(struct got_fileindex *fileindex)
342 {
343 	struct got_fileindex_entry *ie;
344 
345 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
346 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
347 		got_fileindex_entry_free(ie);
348 	}
349 	free(fileindex);
350 }
351 
352 static const struct got_error *
353 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
354 {
355 	size_t n;
356 
357 	val = htobe64(val);
358 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
359 	n = fwrite(&val, 1, sizeof(val), outfile);
360 	if (n != sizeof(val))
361 		return got_ferror(outfile, GOT_ERR_IO);
362 	return NULL;
363 }
364 
365 static const struct got_error *
366 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
367 {
368 	size_t n;
369 
370 	val = htobe32(val);
371 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
372 	n = fwrite(&val, 1, sizeof(val), outfile);
373 	if (n != sizeof(val))
374 		return got_ferror(outfile, GOT_ERR_IO);
375 	return NULL;
376 }
377 
378 static const struct got_error *
379 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
380 {
381 	size_t n;
382 
383 	val = htobe16(val);
384 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
385 	n = fwrite(&val, 1, sizeof(val), outfile);
386 	if (n != sizeof(val))
387 		return got_ferror(outfile, GOT_ERR_IO);
388 	return NULL;
389 }
390 
391 static const struct got_error *
392 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
393 {
394 	size_t n, len, pad = 0;
395 	static const uint8_t zero[8] = { 0 };
396 
397 	len = strlen(path);
398 	while ((len + pad) % 8 != 0)
399 		pad++;
400 	if (pad == 0)
401 		pad = 8; /* NUL-terminate */
402 
403 	SHA1Update(ctx, path, len);
404 	n = fwrite(path, 1, len, outfile);
405 	if (n != len)
406 		return got_ferror(outfile, GOT_ERR_IO);
407 	SHA1Update(ctx, zero, pad);
408 	n = fwrite(zero, 1, pad, outfile);
409 	if (n != pad)
410 		return got_ferror(outfile, GOT_ERR_IO);
411 	return NULL;
412 }
413 
414 static const struct got_error *
415 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
416     FILE *outfile)
417 {
418 	const struct got_error *err;
419 	size_t n;
420 	uint32_t stage;
421 
422 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
423 	if (err)
424 		return err;
425 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
426 	if (err)
427 		return err;
428 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
429 	if (err)
430 		return err;
431 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
432 	if (err)
433 		return err;
434 
435 	err = write_fileindex_val32(ctx, ie->uid, outfile);
436 	if (err)
437 		return err;
438 	err = write_fileindex_val32(ctx, ie->gid, outfile);
439 	if (err)
440 		return err;
441 	err = write_fileindex_val32(ctx, ie->size, outfile);
442 	if (err)
443 		return err;
444 
445 	err = write_fileindex_val16(ctx, ie->mode, outfile);
446 	if (err)
447 		return err;
448 
449 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
450 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
451 	if (n != SHA1_DIGEST_LENGTH)
452 		return got_ferror(outfile, GOT_ERR_IO);
453 
454 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
455 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
456 	if (n != SHA1_DIGEST_LENGTH)
457 		return got_ferror(outfile, GOT_ERR_IO);
458 
459 	err = write_fileindex_val32(ctx, ie->flags, outfile);
460 	if (err)
461 		return err;
462 
463 	err = write_fileindex_path(ctx, ie->path, outfile);
464 	if (err)
465 		return err;
466 
467 	stage = got_fileindex_entry_stage_get(ie);
468 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
469 	    stage == GOT_FILEIDX_STAGE_ADD) {
470 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
471 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
472 		    outfile);
473 		if (n != SHA1_DIGEST_LENGTH)
474 			return got_ferror(outfile, GOT_ERR_IO);
475 	}
476 
477 	return NULL;
478 }
479 
480 const struct got_error *
481 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
482 {
483 	const struct got_error *err = NULL;
484 	struct got_fileindex_hdr hdr;
485 	SHA1_CTX ctx;
486 	uint8_t sha1[SHA1_DIGEST_LENGTH];
487 	size_t n;
488 	struct got_fileindex_entry *ie, *tmp;
489 
490 	SHA1Init(&ctx);
491 
492 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
493 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
494 	hdr.nentries = htobe32(fileindex->nentries);
495 
496 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
497 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
498 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
499 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
500 	if (n != sizeof(hdr.signature))
501 		return got_ferror(outfile, GOT_ERR_IO);
502 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
503 	if (n != sizeof(hdr.version))
504 		return got_ferror(outfile, GOT_ERR_IO);
505 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
506 	if (n != sizeof(hdr.nentries))
507 		return got_ferror(outfile, GOT_ERR_IO);
508 
509 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
510 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
511 		ie->flags &= ~GOT_FILEIDX_F_SKIPPED;
512 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
513 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
514 			got_fileindex_entry_free(ie);
515 			continue;
516 		}
517 		err = write_fileindex_entry(&ctx, ie, outfile);
518 		if (err)
519 			return err;
520 	}
521 
522 	SHA1Final(sha1, &ctx);
523 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
524 	if (n != sizeof(sha1))
525 		return got_ferror(outfile, GOT_ERR_IO);
526 
527 	if (fflush(outfile) != 0)
528 		return got_error_from_errno("fflush");
529 
530 	return NULL;
531 }
532 
533 static const struct got_error *
534 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
535 {
536 	size_t n;
537 
538 	n = fread(val, 1, sizeof(*val), infile);
539 	if (n != sizeof(*val))
540 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
541 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
542 	*val = be64toh(*val);
543 	return NULL;
544 }
545 
546 static const struct got_error *
547 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
548 {
549 	size_t n;
550 
551 	n = fread(val, 1, sizeof(*val), infile);
552 	if (n != sizeof(*val))
553 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
554 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
555 	*val = be32toh(*val);
556 	return NULL;
557 }
558 
559 static const struct got_error *
560 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
561 {
562 	size_t n;
563 
564 	n = fread(val, 1, sizeof(*val), infile);
565 	if (n != sizeof(*val))
566 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
567 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
568 	*val = be16toh(*val);
569 	return NULL;
570 }
571 
572 static const struct got_error *
573 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
574 {
575 	const struct got_error *err = NULL;
576 	const size_t chunk_size = 8;
577 	size_t n, len = 0, totlen = chunk_size;
578 
579 	*path = malloc(totlen);
580 	if (*path == NULL)
581 		return got_error_from_errno("malloc");
582 
583 	do {
584 		if (len + chunk_size > totlen) {
585 			char *p = reallocarray(*path, totlen + chunk_size, 1);
586 			if (p == NULL) {
587 				err = got_error_from_errno("reallocarray");
588 				break;
589 			}
590 			totlen += chunk_size;
591 			*path = p;
592 		}
593 		n = fread(*path + len, 1, chunk_size, infile);
594 		if (n != chunk_size) {
595 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
596 			break;
597 		}
598 		SHA1Update(ctx, *path + len, chunk_size);
599 		len += chunk_size;
600 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
601 
602 	if (err) {
603 		free(*path);
604 		*path = NULL;
605 	}
606 	return err;
607 }
608 
609 static const struct got_error *
610 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
611     FILE *infile, uint32_t version)
612 {
613 	const struct got_error *err;
614 	struct got_fileindex_entry *ie;
615 	size_t n;
616 
617 	*iep = NULL;
618 
619 	ie = calloc(1, sizeof(*ie));
620 	if (ie == NULL)
621 		return got_error_from_errno("calloc");
622 
623 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
624 	if (err)
625 		goto done;
626 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
627 	if (err)
628 		goto done;
629 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
630 	if (err)
631 		goto done;
632 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
633 	if (err)
634 		goto done;
635 
636 	err = read_fileindex_val32(&ie->uid, ctx, infile);
637 	if (err)
638 		goto done;
639 	err = read_fileindex_val32(&ie->gid, ctx, infile);
640 	if (err)
641 		goto done;
642 	err = read_fileindex_val32(&ie->size, ctx, infile);
643 	if (err)
644 		goto done;
645 
646 	err = read_fileindex_val16(&ie->mode, ctx, infile);
647 	if (err)
648 		goto done;
649 
650 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
651 	if (n != SHA1_DIGEST_LENGTH) {
652 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
653 		goto done;
654 	}
655 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
656 
657 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
658 	if (n != SHA1_DIGEST_LENGTH) {
659 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
660 		goto done;
661 	}
662 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
663 
664 	err = read_fileindex_val32(&ie->flags, ctx, infile);
665 	if (err)
666 		goto done;
667 
668 	err = read_fileindex_path(&ie->path, ctx, infile);
669 	if (err)
670 		goto done;
671 
672 	if (version >= 2) {
673 		uint32_t stage = got_fileindex_entry_stage_get(ie);
674 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
675 		    stage == GOT_FILEIDX_STAGE_ADD) {
676 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
677 			    infile);
678 			if (n != SHA1_DIGEST_LENGTH) {
679 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
680 				goto done;
681 			}
682 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
683 		}
684 	} else {
685 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
686 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
687 	}
688 
689 done:
690 	if (err)
691 		got_fileindex_entry_free(ie);
692 	else
693 		*iep = ie;
694 	return err;
695 }
696 
697 const struct got_error *
698 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
699 {
700 	const struct got_error *err = NULL;
701 	struct got_fileindex_hdr hdr;
702 	SHA1_CTX ctx;
703 	struct got_fileindex_entry *ie;
704 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
705 	uint8_t sha1[SHA1_DIGEST_LENGTH];
706 	size_t n;
707 	int i;
708 
709 	SHA1Init(&ctx);
710 
711 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
712 	if (n != sizeof(hdr.signature)) {
713 		if (n == 0) /* EOF */
714 			return NULL;
715 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
716 	}
717 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
718 	if (n != sizeof(hdr.version)) {
719 		if (n == 0) /* EOF */
720 			return NULL;
721 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
722 	}
723 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
724 	if (n != sizeof(hdr.nentries)) {
725 		if (n == 0) /* EOF */
726 			return NULL;
727 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
728 	}
729 
730 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
731 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
732 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
733 
734 	hdr.signature = be32toh(hdr.signature);
735 	hdr.version = be32toh(hdr.version);
736 	hdr.nentries = be32toh(hdr.nentries);
737 
738 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
739 		return got_error(GOT_ERR_FILEIDX_SIG);
740 	if (hdr.version > GOT_FILE_INDEX_VERSION)
741 		return got_error(GOT_ERR_FILEIDX_VER);
742 
743 	for (i = 0; i < hdr.nentries; i++) {
744 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
745 		if (err)
746 			return err;
747 		err = add_entry(fileindex, ie);
748 		if (err)
749 			return err;
750 	}
751 
752 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
753 	if (n != sizeof(sha1_expected))
754 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
755 	SHA1Final(sha1, &ctx);
756 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
757 		return got_error(GOT_ERR_FILEIDX_CSUM);
758 
759 	return NULL;
760 }
761 
762 static struct got_fileindex_entry *
763 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
764 {
765 	struct got_fileindex_entry *next;
766 
767 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
768 
769 	/* Skip entries which were added or removed by diff callbacks. */
770 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
771 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
772 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
773 
774 	return next;
775 }
776 
777 static const struct got_error *
778 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
779     struct got_tree_object *tree, const char *, const char *,
780     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
781 
782 static const struct got_error *
783 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
784     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
785     const char *path, const char *entry_name, struct got_repository *repo,
786     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
787 {
788 	const struct got_error *err = NULL;
789 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
790 
791 	if (!got_object_tree_entry_is_submodule(te) &&
792 	    S_ISDIR(got_tree_entry_get_mode(te))) {
793 		char *subpath;
794 		struct got_tree_object *subtree;
795 
796 		if (asprintf(&subpath, "%s%s%s", path,
797 		    path[0] == '\0' ? "" : "/",
798 		    got_tree_entry_get_name(te)) == -1)
799 			return got_error_from_errno("asprintf");
800 
801 		err = got_object_open_as_tree(&subtree, repo,
802 		    got_tree_entry_get_id(te));
803 		if (err) {
804 			free(subpath);
805 			return err;
806 		}
807 
808 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
809 		    entry_name, repo, cb, cb_arg);
810 		free(subpath);
811 		got_object_tree_close(subtree);
812 		if (err)
813 			return err;
814 	}
815 
816 	(*tidx)++;
817 	*next = got_object_tree_get_entry(tree, *tidx);
818 	return NULL;
819 }
820 
821 static const struct got_error *
822 diff_fileindex_tree(struct got_fileindex *fileindex,
823     struct got_fileindex_entry **ie, struct got_tree_object *tree,
824     const char *path, const char *entry_name, struct got_repository *repo,
825     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
826 {
827 	const struct got_error *err = NULL;
828 	struct got_tree_entry *te = NULL;
829 	size_t path_len = strlen(path);
830 	struct got_fileindex_entry *next;
831 	int tidx = 0;
832 
833 	te = got_object_tree_get_entry(tree, tidx);
834 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
835 		if (te && *ie) {
836 			char *te_path;
837 			const char *te_name = got_tree_entry_get_name(te);
838 			int cmp;
839 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
840 				err = got_error_from_errno("asprintf");
841 				break;
842 			}
843 			cmp = got_path_cmp((*ie)->path, te_path,
844 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
845 			free(te_path);
846 			if (cmp == 0) {
847 				if (got_path_is_child((*ie)->path, path,
848 				    path_len) &&
849 				    !got_object_tree_entry_is_submodule(te) &&
850 				    (entry_name == NULL ||
851 				    strcmp(te_name, entry_name) == 0)) {
852 					err = cb->diff_old_new(cb_arg, *ie, te,
853 					    path);
854 					if (err || entry_name)
855 						break;
856 				}
857 				*ie = walk_fileindex(fileindex, *ie);
858 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
859 				    path, entry_name, repo, cb, cb_arg);
860 			} else if (cmp < 0) {
861 				next = walk_fileindex(fileindex, *ie);
862 				if (got_path_is_child((*ie)->path, path,
863 				    path_len) && entry_name == NULL) {
864 					err = cb->diff_old(cb_arg, *ie, path);
865 					if (err || entry_name)
866 						break;
867 				}
868 				*ie = next;
869 			} else {
870 				if ((entry_name == NULL ||
871 				    strcmp(te_name, entry_name) == 0)) {
872 					err = cb->diff_new(cb_arg, te, path);
873 					if (err || entry_name)
874 						break;
875 				}
876 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
877 				    path, entry_name, repo, cb, cb_arg);
878 			}
879 			if (err)
880 				break;
881 		} else if (*ie) {
882 			next = walk_fileindex(fileindex, *ie);
883 			if (got_path_is_child((*ie)->path, path, path_len) &&
884 			    (entry_name == NULL ||
885 			    (te && strcmp(got_tree_entry_get_name(te),
886 			    entry_name) == 0))) {
887 				err = cb->diff_old(cb_arg, *ie, path);
888 				if (err || entry_name)
889 					break;
890 			}
891 			*ie = next;
892 		} else if (te) {
893 			if (!got_object_tree_entry_is_submodule(te) &&
894 			    (entry_name == NULL ||
895 			    strcmp(got_tree_entry_get_name(te), entry_name)
896 			    == 0)) {
897 				err = cb->diff_new(cb_arg, te, path);
898 				if (err || entry_name)
899 					break;
900 			}
901 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
902 			    entry_name, repo, cb, cb_arg);
903 			if (err)
904 				break;
905 		}
906 	}
907 
908 	return err;
909 }
910 
911 const struct got_error *
912 got_fileindex_diff_tree(struct got_fileindex *fileindex,
913     struct got_tree_object *tree, const char *path, const char *entry_name,
914     struct got_repository *repo,
915     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
916 {
917 	struct got_fileindex_entry *ie;
918 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
919 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
920 		ie = walk_fileindex(fileindex, ie);
921 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
922 	    cb, cb_arg);
923 }
924 
925 static const struct got_error *
926 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
927     struct got_pathlist_head *, int, const char *, const char *,
928     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
929 
930 static const struct got_error *
931 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
932 {
933 	const struct got_error *err = NULL;
934 	struct got_pathlist_entry *new = NULL;
935 	struct dirent *dep = NULL;
936 	struct dirent *de = NULL;
937 
938 	for (;;) {
939 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
940 		if (de == NULL) {
941 			err = got_error_from_errno("malloc");
942 			break;
943 		}
944 
945 		if (readdir_r(dir, de, &dep) != 0) {
946 			err = got_error_from_errno("readdir_r");
947 			free(de);
948 			break;
949 		}
950 		if (dep == NULL) {
951 			free(de);
952 			break;
953 		}
954 
955 		if (strcmp(de->d_name, ".") == 0 ||
956 		    strcmp(de->d_name, "..") == 0 ||
957 		    (path[0] == '\0' &&
958 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
959 			free(de);
960 			continue;
961 		}
962 
963 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
964 		if (err) {
965 			free(de);
966 			break;
967 		}
968 		if (new == NULL) {
969 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
970 			free(de);
971 			break;
972 		}
973 	}
974 
975 	return err;
976 }
977 
978 void
979 free_dirlist(struct got_pathlist_head *dirlist)
980 {
981 	struct got_pathlist_entry *dle;
982 
983 	TAILQ_FOREACH(dle, dirlist, entry)
984 		free(dle->data);
985 	got_pathlist_free(dirlist);
986 }
987 
988 static int
989 have_tracked_file_in_dir(struct got_fileindex *fileindex, const char *path)
990 {
991 	struct got_fileindex_entry *ie;
992 	size_t path_len = strlen(path);
993 	int cmp;
994 
995 	ie = RB_ROOT(&fileindex->entries);
996 	while (ie) {
997 		if (got_path_is_child(ie->path, path, path_len))
998 			return 1;
999 		cmp = got_path_cmp(path, ie->path, path_len,
1000 		    got_fileindex_entry_path_len(ie));
1001 		if (cmp < 0)
1002 			ie = RB_LEFT(ie, entry);
1003 		else if (cmp > 0)
1004 			ie = RB_RIGHT(ie, entry);
1005 		else
1006 			break;
1007 	}
1008 
1009 	return 0;
1010 }
1011 
1012 static const struct got_error *
1013 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
1014     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
1015     const char *path, const char *rootpath, struct got_repository *repo,
1016     int ignore, struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1017 {
1018 	const struct got_error *err = NULL;
1019 	struct dirent *de = dle->data;
1020 	DIR *subdir = NULL;
1021 	int subdirfd = -1;
1022 	int type;
1023 
1024 	*next = NULL;
1025 
1026 	if (de->d_type == DT_UNKNOWN) {
1027 		/* Occurs on NFS mounts without "readdir plus" RPC. */
1028 		char *dir_path;
1029 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
1030 			return got_error_from_errno("asprintf");
1031 		err = got_path_dirent_type(&type, dir_path, de);
1032 		free(dir_path);
1033 		if (err)
1034 			return err;
1035 	} else
1036 		type = de->d_type;
1037 
1038 	/* Must traverse ignored directories if they contain tracked files. */
1039 	if (type == DT_DIR && ignore &&
1040 	    have_tracked_file_in_dir(fileindex, path))
1041 		ignore = 0;
1042 
1043 	if (type == DT_DIR && !ignore) {
1044 		char *subpath;
1045 		char *subdirpath;
1046 		struct got_pathlist_head subdirlist;
1047 
1048 		TAILQ_INIT(&subdirlist);
1049 
1050 		if (asprintf(&subpath, "%s%s%s", path,
1051 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1052 			return got_error_from_errno("asprintf");
1053 
1054 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
1055 			free(subpath);
1056 			return got_error_from_errno("asprintf");
1057 		}
1058 
1059 		subdirfd = openat(fd, de->d_name,
1060 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY);
1061 		if (subdirfd == -1) {
1062 			if (errno == EACCES) {
1063 				*next = TAILQ_NEXT(dle, entry);
1064 				return NULL;
1065 			}
1066 			err = got_error_from_errno2("openat", subdirpath);
1067 			free(subpath);
1068 			free(subdirpath);
1069 			return err;
1070 		}
1071 
1072 		subdir = fdopendir(subdirfd);
1073 		if (subdir == NULL)
1074 			return got_error_from_errno2("fdopendir", path);
1075 		subdirfd = -1;
1076 		err = read_dirlist(&subdirlist, subdir, subdirpath);
1077 		if (err) {
1078 			free(subpath);
1079 			free(subdirpath);
1080 			closedir(subdir);
1081 			return err;
1082 		}
1083 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1084 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1085 		if (subdir && closedir(subdir) == -1 && err == NULL)
1086 			err = got_error_from_errno2("closedir", subdirpath);
1087 		free(subpath);
1088 		free(subdirpath);
1089 		free_dirlist(&subdirlist);
1090 		if (err)
1091 			return err;
1092 	}
1093 
1094 	*next = TAILQ_NEXT(dle, entry);
1095 	return NULL;
1096 }
1097 
1098 static const struct got_error *
1099 diff_fileindex_dir(struct got_fileindex *fileindex,
1100     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1101     int dirfd, const char *rootpath, const char *path,
1102     struct got_repository *repo,
1103     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1104 {
1105 	const struct got_error *err = NULL;
1106 	struct dirent *de = NULL;
1107 	size_t path_len = strlen(path);
1108 	struct got_pathlist_entry *dle;
1109 	int ignore;
1110 
1111 	if (cb->diff_traverse) {
1112 		err = cb->diff_traverse(cb_arg, path, dirfd);
1113 		if (err)
1114 			return err;
1115 	}
1116 
1117 	dle = TAILQ_FIRST(dirlist);
1118 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1119 		if (dle && *ie) {
1120 			char *de_path;
1121 			int cmp;
1122 			de = dle->data;
1123 			if (asprintf(&de_path, "%s/%s", path,
1124 			    de->d_name) == -1) {
1125 				err = got_error_from_errno("asprintf");
1126 				break;
1127 			}
1128 			cmp = got_path_cmp((*ie)->path, de_path,
1129 			    got_fileindex_entry_path_len(*ie),
1130 			    strlen(path) + 1 + strlen(de->d_name));
1131 			free(de_path);
1132 			if (cmp == 0) {
1133 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1134 				    dirfd);
1135 				if (err)
1136 					break;
1137 				*ie = walk_fileindex(fileindex, *ie);
1138 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1139 				    path, rootpath, repo, 0, cb, cb_arg);
1140 			} else if (cmp < 0 ) {
1141 				err = cb->diff_old(cb_arg, *ie, path);
1142 				if (err)
1143 					break;
1144 				*ie = walk_fileindex(fileindex, *ie);
1145 			} else {
1146 				err = cb->diff_new(&ignore, cb_arg, de, path,
1147 				    dirfd);
1148 				if (err)
1149 					break;
1150 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1151 				    path, rootpath, repo, ignore, cb, cb_arg);
1152 			}
1153 			if (err)
1154 				break;
1155 		} else if (*ie) {
1156 			err = cb->diff_old(cb_arg, *ie, path);
1157 			if (err)
1158 				break;
1159 			*ie = walk_fileindex(fileindex, *ie);
1160 		} else if (dle) {
1161 			de = dle->data;
1162 			err = cb->diff_new(&ignore, cb_arg, de, path, dirfd);
1163 			if (err)
1164 				break;
1165 			err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1166 			    rootpath, repo, ignore, cb, cb_arg);
1167 			if (err)
1168 				break;
1169 		}
1170 	}
1171 
1172 	return err;
1173 }
1174 
1175 const struct got_error *
1176 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1177     const char *rootpath, const char *path, struct got_repository *repo,
1178     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1179 {
1180 	const struct got_error *err;
1181 	struct got_fileindex_entry *ie;
1182 	struct got_pathlist_head dirlist;
1183 	int fd2;
1184 	DIR *dir;
1185 
1186 	TAILQ_INIT(&dirlist);
1187 
1188 	/* 
1189 	 * Duplicate the file descriptor so we can call closedir() below
1190 	 * without closing the file descriptor passed in by our caller.
1191 	 */
1192 	fd2 = dup(fd);
1193 	if (fd2 == -1)
1194 		return got_error_from_errno2("dup", path);
1195 	if (lseek(fd2, 0, SEEK_SET) == -1) {
1196 		err = got_error_from_errno2("lseek", path);
1197 		close(fd2);
1198 		return err;
1199 	}
1200 	dir = fdopendir(fd2);
1201 	if (dir == NULL) {
1202 		err = got_error_from_errno2("fdopendir", path);
1203 		close(fd2);
1204 		return err;
1205 	}
1206 	err = read_dirlist(&dirlist, dir, path);
1207 	if (err) {
1208 		closedir(dir);
1209 		return err;
1210 	}
1211 
1212 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1213 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1214 		ie = walk_fileindex(fileindex, ie);
1215 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1216 	    rootpath, path, repo, cb, cb_arg);
1217 
1218 	if (closedir(dir) == -1 && err == NULL)
1219 		err = got_error_from_errno2("closedir", path);
1220 	free_dirlist(&dirlist);
1221 	return err;
1222 }
1223 
1224 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);